.
We weigh AI by the measure of language, and build it on language's terms.
Language is both the material and the arena of generative AI: models learn from it, are addressed through it, and have their output weighed by its standards. As the Kingdom pursues global leadership in data and AI under Vision 2030, and seeks to enable the use of Arabic across the many domains of life, as the National Policy for the Arabic Language sets out, the need grows for people who can weigh these technologies by the measure of language: faithfulness in translation, soundness in Arabic, suitability in teaching, and fairness in how languages and cultures are represented.
This is where the Artificial Intelligence and Linguistic Technologies Lab at the College of Language Sciences, King Saud University, begins. It draws on the College's seven departments, where eleven languages are taught, and on long-standing expertise in translation, linguistics, language teaching, and literature. The Lab does two things: it evaluates language AI technologies with a scientific methodology whose results are published and built upon, and it contributes to developing language applications whose quality is judged by specialists rather than by automatic metrics alone. It carries out its mission through teaching, research, community service, and partnerships, and it is the College's link to the University's Artificial Intelligence Office in carrying out the College's AI plan.
The Lab has published guides in Arabic and English on using generative AI in teaching and assessment, presented the prototype of Yusr, an AI assistant that makes Arabic educational content accessible to people with disabilities and learning difficulties, and continues its research and training in the evaluation of AI translation.
Vision:
To be the reference others turn to on the quality of language AI technologies, and a partner in their development: serving Arabic and the world's languages, and contributing to building a knowledge society.
Mission:
To harness the College's expertise in languages and translation to evaluate and develop language AI technologies: through research that sets the standards for their evaluation, teaching that prepares national talent to use them with awareness and responsibility, and local and international partnerships that serve society and carry this expertise into education, translation, and technology.
Key Areas:
The Lab studies AI in three roles, with Arabic present in all of them:
1. AI as translator: the quality of machine and generative translation between Arabic and the College's languages, error analysis, post-editing, and measurement, in legal, media, and literary texts.
2. AI as teacher: the soundness of its outputs in language learning, teaching, and assessment, from feedback on writing to the preparation of teaching materials, while safeguarding academic integrity and setting standards for responsible use.
3. AI as a mirror of languages and cultures: how Arabic and its dialects, literary traditions, and less-represented cultures appear in language models and their training corpora, and the bias or erasure that can creep in.
Objectives:
1. Standards. Establish standards for evaluating AI outputs in Arabic and the College's languages: a unified methodology, an error taxonomy that respects the features of Arabic, and test corpora that are published and serve as a reference.
2. Teaching. Integrate AI into the College's academic programs, in teaching and assessment, through guides, courses, and hands-on workshops, and prepare faculty and students to use it with awareness and responsibility.
3. Research. Advance research on language AI and on how it represents languages and cultures, and involve graduate students in its projects to prepare a new generation of researchers.
4. Applications. Develop intelligent language applications and tools that serve Arabic and the College's languages, with technology partners, tested against the Lab's standards before they reach their users.
5. Partnerships. Build local and international partnerships with research and technology organizations that draw on the College's evaluation expertise and return joint research, resources, and self-generated revenue to the College.
6. Outreach. Share the Lab's expertise with the College and the wider community through scholarly lectures and specialist meetings with industry and business.
